preCICE watchpoint for CalculiX reports wrong data for restarted simulations

Hi. For the FSI validation case of cylinder with trailing flap, I made a comparison of displacements Ux Uy at point A obtained from Calculix node output file as well as preCICE watchpoint file. There seems to be a minor deviation between the data reported by Calculix and preCICE. Thus, I would like to know what can be the reason for this deviation, and how the displacement data is generated by preCICE for the watchpoint file. Is the methodology used to obtain interface displacement for coupling process identical to the one used for printing the watchpoint file?

This post is an update regarding the issue of deviation between Calculix displacement and preCICE watchpoint data that was posted by me on Gitter, which is also referenced above. On further investigation for other time intervals, the deviation was worse (see link ). The results in the above link were for a coupled simulation that was restarted from 3.7675[s].

Also, I attempted a coupled simulation starting from t = 0[s]. Unlike the previous attempt, there appears to be no deviation in the duration of interest, with the exception of phase shift identical to the original discussion I posted in Gitter (see link ).

Since the Calculix data for restarted simulation as well as the original simulation is identical, it leads me to suspect that in regions of significant displacement, the preCICE watchpoint report seems to be faulty for a restarted simulation. Also for any simulation, there is always a phase difference of 1 timestep (here dt = 0.0005[s]) between the Calculix and preCICE data.

@ksripathy are you defining the watchpoint on a provided or on a received-from mesh? Since preCICE v2.1.0 it will only be possible to define them on provided meshes and maybe your problems is side-cured by this.

How does your config look like?


When restarting a simulation, CalculiX uses the original coordinates from the undeformed geometry as its mesh. These original coordinates + the displacements from the previous simulations last time step is sent to preCICE for the mapping as the new simulation starts from the previously deformed geometry. I assume that this is where the discrepancy is occurring.

When plotting the displacements in CalculiX, you are plotting the real displacements. But in preCICE, the displacement might actually be displacement - (Original coordinates - initial displacement). This is my best guess and we will try to check if this is the problem. Are you using absolute displacements? Is the problem better when using displacementDeltas?